![Small delivery robot on a London pavement — AI-generated symbolic image, not a Coco vehicle [Image content created with AI]](https://alpha-bionic.info/wp-content/uploads/2026/10/deliveroo-coco-symbolbild-20261006-1.webp "Deliveroo plans robot deliveries in five UK locations 1 [Image content created with AI]")
Deliveroo and Coco Robotics plan to introduce pavement delivery robots in the UK in October 2026. The first location is set to be Canary Wharf in London, with Milton Keynes, Leeds, Stockton-on-Tees and Nottingham to follow. The companies announced the partnership on 5 October. This is a rollout plan, not evidence that a nationwide robot service is already operating.
Image: AI-generated symbolic illustration of a pavement delivery robot; not a Coco vehicle or a photograph of the planned UK service.
In this article
What has actually been announced?
The exclusive partnership would connect Coco’s small electric robots to Deliveroo’s ordering platform for suitable short-distance deliveries. The companies name restaurants and shops in Wood Wharf, including Patty&Bun, Signorelli and Beam, as initial participants. More merchants and a wider service area are expected later. Some prospective partners in Milton Keynes are also named. Yet the announcement does not specify a launch date for each of the five locations, the number of vehicles, or the share of orders robots may handle.
That distinction matters. An agreement between platforms, a local launch, a completed customer delivery and a dependable everyday service are four separate milestones. The announcement says Canary Wharf should begin this October; subsequent towns and cities are slated for the coming weeks and months. Readers should not confuse that timetable with verified operations across the UK.
Why start in Canary Wharf?
Canary Wharf offers a concentrated mix of restaurants, offices and short journeys. Its estate operator has publicly welcomed the project. Such a setting can make an early deployment easier: predictable routes, nearby customers and local cooperation reduce some of the challenges found elsewhere. It also limits what the first results can prove. A robot that performs well on a managed estate may face different conditions on narrow residential pavements, at difficult crossings or in districts with dispersed customers.
The real test is not a smooth promotional journey. It is whether robots can complete enough useful trips throughout an ordinary week while remaining safe and accessible for everyone who uses the pavement.
How the delivery robots work
According to Deliveroo, Coco’s vehicles travel at walking pace and use onboard cameras, sensors and navigation software to respond to their surroundings. Environmental processing is said to happen locally on the vehicle rather than relying on an uninterrupted cloud connection. Rules for speed and routing can be adapted to a location, while trained operators supervise operations. In this context, “autonomous” should not be read as “never needs human oversight.”
The announcement leaves important operational metrics unanswered. How often must a supervisor intervene? How does the system handle roadworks, dropped kerbs, crowded footways or a route blocked by a parked bicycle? How long does a complete delivery take once loading, travel and customer collection are included? Without published figures, neither reliability nor the economic advantage over existing couriers can be assessed independently.
A complement to riders, not a wholesale replacement
Deliveroo says most orders will continue to be fulfilled by its rider network. Robots are intended for a subset of shorter trips in suitable urban areas. That makes sense technically: a low-speed pavement vehicle has limited cargo capacity, range and route choice. It cannot take every order a cyclist or motorbike courier could serve.
For restaurants, another delivery option could help when demand peaks, provided the service is available and predictable. But robots also require charging, maintenance, cleaning, depot space, software support and people who can recover stranded units. Coco and Deliveroo anticipate UK operational infrastructure and skilled jobs, yet offer no public cost-per-delivery comparison. An economically attractive service depends on the number of suitable orders per robot and how many vehicles one operator can responsibly supervise.
The pavement is shared space
The central public question is not simply whether a robot can avoid a collision. It must behave predictably around pedestrians, wheelchair users, children and people with visual impairments. Coco points to a partnership with BlindSquare under which robots may provide live information about pavement hazards to a navigation app. This could be useful, but the announcement does not include an independent evaluation of the feature in the planned UK locations.
Likewise, an electrically powered trip has no tailpipe emissions, but that is not the same as a zero-impact delivery chain. Vehicle production, charging, depots and support journeys all matter. A credible environmental comparison would measure effects per completed order against the alternatives available in that particular area.
What would count as progress?
Alpha Bionic’s key distinction is between a promising deployment design and proven daily utility. The UK pilot becomes meaningful when customers can place real orders, completion and intervention rates are reported, and the service remains available beyond its launch phase. Safety and accessibility outcomes matter as much as speed. Until then, the accurate headline is that Deliveroo and Coco plan robot deliveries in five British locations. The operational evidence still has to arrive.
Sources
- Deliveroo: partnership announcement and planned locations, 5 October 2026.
- The Grocer: report on the planned rollout, 5 October 2026.
- The Next Web: coverage of the UK expansion, 5 October 2026.